J Am Soc Nephrol 13: 1918–1927, 2002

Traditional Cardiovascular Disease Risk Factors in Dialysis

Patients Compared with the General Population: The CHOICE
Study
J. CRAIG LONGENECKER,* JOSEF CORESH,* NEIL R. POWE,*
ANDREW S. LEVEY,† NANCY E. FINK,* ALICE MARTIN,† and MICHAEL J. KLAG*
*Johns Hopkins University, Baltimore, Maryland; and †New England Medical Center, Boston, Massachusetts.

Abstract. Although atherosclerotic cardiovascular disease (AS- revascularization procedure, stroke, carotid endarterectomy,
CVD) risk in end-stage renal disease (ESRD) is 5 to 30 times and amputation in CHOICE; and as myocardial infarction and
that of the general population, few data exist comparing AS- stroke in NHANES III), the prevalence of diabetes, hyperten-
CVD risk factors among new dialysis patients to the general sion, left ventricular hypertrophy by EKG, low physical activ-
population. This cross-sectional study of 1041 dialysis patients ity, low HDL cholesterol, and hypertriglyceridemia were still
describes the prevalence of ASCVD risk factors at the begin- more common in CHOICE participants. Smoking, obesity,
ning of ESRD compared with estimates of ASCVD risk factors hypercholesterolemia, and high LDL cholesterol, however,
in the adult US population derived from the Third National were less common in CHOICE than NHANES III participants.
Health and Nutrition Examination (NHANES III). CHOICE The projected 5-yr ASCVD risk based on the Framingham
Study participants had a high prevalence of diabetes (54%), Risk Equation among those older than 40 yr without ASCVD
hypertension (96%), left ventricular hypertrophy by electrocar- was higher in CHOICE Study participants (13%) than in the
diogram (EKG) criteria (22%), low physical activity (80%), NHANES III participants (6%). In summary, many ASCVD
hypertriglyceridemia (36%), and low HDL cholesterol (33%). risk factors are more prevalent in ESRD than in the general
CHOICE participants were more likely to be older, black, and population and may explain some, but probably not all, of the
male than NHANES III participants. After adjustment for age, increased ASCVD risk in ESRD.
race, gender, and ASCVD (defined as myocardial infarction,

Atherosclerotic cardiovascular disease (ASCVD) accounts for timate the presence and effect of risk factors because those
approximately half of deaths in end-stage renal disease (ESRD) with the highest degree of ASCVD risk tend to die sooner and
and contributes to the extraordinarily high total annual mortal- are not included in a prevalent study population (i.e., survival
ity of 23% observed in such patients (1). The incidence of bias), an effect diminished but not eliminated by cross-sec-
myocardial infarction (MI) and stroke in the dialysis popula- tional studies of incident dialysis patients.
tion is 5- to 15-fold higher in ESRD (2), and cardiovascular Relatively few nationally representative studies (4,9,18 –20)
mortality is 10- to 30-fold higher (3) than that seen in the have described selected ASCVD risk factors among incident
general population (4 – 6), This increased risk is only partially dialysis patients. Several other regional (21,22) and local
explained by a high prevalence of ASCVD (2,4,7–9) and (23,24) studies of incident patients have also been reported.
traditional ASCVD risk factors (10) at the initiation of dialysis None of these studies, however, compares ASCVD risk factor
(3,11,12). prevalence in the incident dialysis population with the general
The Special Report from the National Kidney Foundation population.
Task Force on Cardiovascular Disease (13) called for further This report presents the prevalence of ASCVD risk factors in
studies of ASCVD and its risk factors in ESRD patients. Most the Choices for Healthy Outcomes in Caring for ESRD
previous studies of ASCVD risk factors have investigated (CHOICE) Study, a national study of incident dialysis patients
prevalent ESRD patients (14 –17). Such studies may underes- (25), compared with estimates for the general population de-
rived from the Third National Health and Nutrition Examina-
tion Survey (NHANES III). Because age, gender, race, and the
presence of ASCVD are strongly associated both with ASCVD
Received September 7, 2001. Accepted April 6, 2002. risk factors and ESRD, the NHANES estimates used for the
Correspondence to Dr. J. Craig Longenecker, Johns Hopkins University School of
Medicine, Suite 2-637, 2024 E Monument Street, Baltimore, MD 21205. Phone:
comparison are adjusted to the age, gender, race, and ASCVD
410-614-6928; Fax: 410-955-0476; E-mail: [email protected] distribution of the CHOICE cohort. A second analysis uses the
1046-6673/1307-1918 Framingham risk equation to estimate the 1- and 5-yr ASCVD
Journal of the American Society of Nephrology risk among those without prevalent ASCVD and compares the
Copyright © 2002 by the American Society of Nephrology derived ASCVD risk estimates from the CHOICE cohort with
DOI: 10.1097/01.ASN.0000019641.41496.1E those of the NHANES III study population.
J Am Soc Nephrol 13: 1918–1927, 2002 Cardiovascular Risk Factors in ESRD 1919

Materials and Methods 971 (93%), 946 (91%), 943 (91%), and 653 (63%) of the cohort.
Study Design and Research Population When risk factors were analyzed separately, all participants with
This cross-sectional study is derived from the baseline data of information were included. When risk factors were combined for
CHOICE, a prospective cohort study of incident dialysis patients analyses (e.g., Table 4), only participants with complete information
initiated in 1995 to investigate treatment choices and outcomes of on all variables were included in analyses.
dialysis care. Eligibility criteria for enrollment into CHOICE included Specimen Bank and Laboratory Assays. Nonfasting venous
initiation of chronic outpatient dialysis in the preceding 3 mo, ability serum specimens are collected at the DCI dialysis facilities just before
to provide informed consent for participation, age older than 17 yr, a dialysis session. Specimens are spun at 2500 to 3000 rpm and
and ability to speak English or Spanish. The Johns Hopkins University filtered on site within 45 min of phlebotomy and sent overnight to the
School of Medicine Institutional Review Board and the review boards DCI Central Laboratory (Nashville, TN), where they are stored at
for the clinical centers approved the study protocol. ⫺80°C. More than 95% of samples are frozen within 48 h of veni-
From October 1995 to June 1998, 1041 participants from 19 states puncture. The CHOICE cohort enrolled incident dialysis patients, but
were enrolled at 81 dialysis clinics associated with Dialysis Clinic Inc. serologic parameters may be highly variable at the initiation of dial-
(DCI, Nashville, TN; n ⫽ 923), New Haven CAPD (New Haven, CT; ysis and may not reflect an individual’s long-term level because of
n ⫽ 86), or Saint Raphael’s Hospital (New Haven, CT; n ⫽ 32). A changes in dialysis dose and clinical status. To provide a more stable
specimen bank was established to store blood samples from the DCI estimate of an individual’s level of serologic markers, samples drawn
enrollees, and specimens were obtained for 898 (97.3%) of the DCI at approximately 3 mo after enrollment were used. The median time
participants, allowing for measurement of complete lipid profiles in from enrollment to collection was 2.8 mo, with 95% of samples
this subgroup. In addition, blood test results obtained from routine obtained within 4.8 mo. The median time from first dialysis to serum
medical care were available for all 1041 participants. Enrollment collection was 4.4 mo, with 95% of samples obtained within 7 mo.
occurred a median of 45 d after first dialysis (98% within 4 mo). Laboratories performing all assays were blinded to all clinical infor-
Comorbidity data from the Medical Evidence Report (Form 2728 of mation, including age, race, gender, and comorbid conditions.
the US Renal Data System [USRDS]) were used to compare charac- Colorimetric methods that used an Olympus (Hamburg, Germany)
teristics of the CHOICE cohort to the characteristics of all incident autoanalyzer were used to determine total cholesterol (coefficient of
dialysis patients in the United States in 1997 (the midpoint of recruit- variation [CV], 5.3%), HDL cholesterol (CV, 9.6%), and triglyceride
ment). Although these data have been shown to underestimate the (CV, 12.3%) levels (all CV values were determined by blinded split
prevalence of comorbid conditions in incident dialysis patients (26), samples; n ⫽ 39). The Friedewald formula was used to calculate LDL
they provide an identical data source for comparisons between cholesterol for those with triglycerides ⬍400 mg/dl. Apolipopro-
CHOICE and the US dialysis population. tein-A1 (CV, 12.3%) and apolipoprotein-B (CV, 9.5%) were mea-
sured via immunonephelometric methods with a Dade-Behring (Mar-
burg, Germany) autoanalyzer. Of the 898 specimen bank participants
Data Collection with serum available, total cholesterol, triglycerides, and HDL cho-
CHOICE Clinical Data. Age, race, gender, physical activity, lesterol data were available for 862 individuals (96%). For calculation
and tobacco use history were obtained via a questionnaire adminis- of the Framingham risk equation, the baseline total cholesterol ob-
tered to the patient. Weight, height and pre- and postdialysis session tained for routine care was used to fill in missing data for those not
BP were obtained from review of the patients’ medical records. able to participate in the specimen bank (n ⫽ 19).
Prevalent ASCVD, diabetes, hypertension, and left ventricular hyper- NHANES III Data. To obtain population-based estimates of
trophy (LVH) by electrocardiogram (EKG) criteria were determined ASCVD risk factors, we used data from NHANES III (30 –33). The
at enrollment on the bases of review of all history and physical data, sample design used complex, multistage, clustered samples of civil-
discharge summaries, progress notes, medication records, EKG, and ian, noninstitutionalized populations. A total of 20,050 adults were
problem lists from the dialysis clinic chart. All records were ab- interviewed and examined. Of these, we analyzed 19,753 who had
stracted by two experienced dialysis research nurses at the CHOICE complete data on age, gender, race, and history of MI and cerebro-
Comorbidity Assessment Center (New England Medical Center, Bos- vascular accident (the NHANES definition of ASCVD). Of these,
ton, MA). Mention of a condition (past or present) in the medical 19,395 had BP measured; 17,848 had self-reported diabetes, smoking,
record was sufficient for positive coding. body mass index, physical activity, and congestive heart failure data;
In the CHOICE study, ASCVD was defined as a history of MI, 8436 (those older than 40 only) had EKG evaluated for evidence of
coronary artery bypass or angioplasty, carotid endarterectomy, stroke, LVH; and 16,870 had cholesterol, triglycerides, and HDL cholesterol
peripheral bypass, peripheral angioplasty, or amputation. The defini- measured. To correspond to the CHOICE questionnaire, the frequen-
tion of diabetes included both type 1 and type 2 diabetes. Current cies of all 5.0 MET or greater activities were tabulated and combined.
physical activity was determined by two questions: “At least once a
week, do you engage in any regular exercise such as brisk walking, Statistical Analyses
jogging, bicycling, etc., long enough to work up a sweat?” and “If so, Statistical analyses were performed with STATA (version 6.0).
how many times per week?” Exercise to perspiration was estimated to Descriptive statistics that used means, medians, proportions, SE, and
be equivalent to a 5.0 metabolic equivalent task (MET) or greater confidence intervals were performed on all variables where appropri-
activity (e.g., stationary bicycling as a conditioning exercise is a 5.0 ate. For the CHOICE data, the exact binomial method was used to
MET activity) (27–29). LVH on EKG was coded positive if the note determine SE for proportions.
“LVH by EKG criteria” was present in chart records or on an EKG The standard NHANES III Mobile Examination Center survey
report. weights were used for survey estimates in the general US population.
Age, race, and gender were available for all CHOICE participants. The NHANES weights were then modified to provide ASCVD risk
Diabetes, ASCVD, and hypertension status was available for 1038 factor and SE estimates adjusted to the age decade, gender, race, and
(99.7%) of 1041 participants. Smoking, body mass index, physical ASCVD distribution of the CHOICE cohort (see Appendix for
activity, BP, and EKG were available, respectively, for 975 (94%), method).
1920 Journal of the American Society of Nephrology J Am Soc Nephrol 13: 1918–1927, 2002

Table 1. Dialysis modality and Medical Evidence Report (Form 2728) characteristics of the CHOICE cohort (recruited
1995–1998), compared with all incident end-stage renal disease (ESRD) patients receiving dialysis
1997 USRDS Incident
CHOICE Cohort
Characteristic Dialysis Patients
(n ⫽ 1041) (n ⫽ 79,102)

Form 2728 demographic data

Mean age, years 60.2
Sex, %
male 54 53
female 46 47
Race, %
white 67 65
black 28 29
other 5 6
Form 2728 comorbidity data
Cause of ESRD, %
diabetes mellitus 47 42
hypertension 17 25
glomerulonephritis 16 9
other 20 23
hypertension, % 74 74
diabetes, % 40 41
insulin dependence, % 29 24
coronary artery disease, % 21 25
myocardial infarction, % 9 9
congestive heart failure, % 25 35
cardiac arrest, % 1.4 1.0
cardiac dysrhythmia, % 5 6.2
pericarditis, % 1.4 1.1
stroke or transitent ischemic attack, % 8 10
peripheral vascular disease, % 13 15
chronic obstructive lung disease, % 6 7
tobacco use, % 8 6
malignancy, % 4 5
Form 2728 Laboratory Data
mean hematocrit (%) 28.6 28.6
mean serum albumin (g/dl) 3.4 3.2
mean serum creatinine (mg/dl) 8.8 8.1
mean serum blood urea nitrogen (mg/dl) 90 92
Modality, %
hemodialysis 73 87a
peritoneal dialysis 27 13a
a
Estimates are based on dialysis modality distribution in the prevalent US Renal Data System (USRDS) population.

The only ASCVD events ascertained by NHANES III included MI overestimate the adjusted NHANES estimates. This is because one
and stroke. Therefore, the adjustment procedure for ASCVD de- would expect ASCVD risk factors to be slightly more prevalent in MI
scribed in the Appendix was predicated on the important assumption or stroke patients than in CABG, PTCA, carotid endarterectomy, or
that the profile of risk factors in NHANES participants with a history peripheral vascular disease patients.
of MI and stroke (which were ascertained) is similar to that of All of the differences between NHANES and CHOICE were highly
NHANES participants with a history of coronary artery bypass graft statistically significant. However, systematic differences in the meth-
(CABG), percutaneous transluminal coronary angioplasty (PTCA), ods used in the two studies may have accounted for some of the
carotid endarterectomy, and peripheral vascular disease (which were differences, and would not have been reflected in P values. We
not ascertained). Although this assumption is most likely not perfect, therefore chose not to present P values for these comparisons.
we believe that any differences that may exist would not result in The Framingham risk equation (34) was used to estimate, at the
significant errors in the adjustment procedure. Furthermore, such a individual level, the theoretical 8-yr cardiovascular risk for the
bias would generally be conservative in nature and would tend to NHANES III and CHOICE populations. The Framingham risk equa-
J Am Soc Nephrol 13: 1918–1927, 2002 Cardiovascular Risk Factors in ESRD 1921

Table 2. Comparison of cardiovascular disease risk factor prevalence adjusted to the CHOICE distribution of age, race,
gender, and prevalent cardiovascular diseasea
CHOICE Cohort (n ⫽ 1041) NHANES III Population (n ⫽ 19,537)
CVD Risk Factors
Unadjusted NHANES Estimates Adjusted
Estimates (SE) III Estimates (SE) to CHOICEb (SE)

Demographics
mean age (yr)b 57.8 (0.5) 43 (0.4) 57.3 (0.4)
gender (% male)b 54 (1.5) 48 (0.4) 54 (1.0)
raceb
white (%) 67 (1.5) 76 (1.2) 67 (1.8)
black (%) 28 (1.4) 11 (0.6) 28 (1.6)
other (%) 5 (0.7) 13 (0.9) 5 (0.5)
Comorbid conditions
diabetes (%) 54 (1.5) 5 (0.2) 15 (0.8)
mean systolic BP (mmHg) 149 (0.6) 122 (0.4) 132 (0.5)
mean diastolic BP (mmHg) 79 (0.3) 74 (0.2) 76 (0.2)
hypertension (%) 96 (0.6) 23 (0.6) 44 (1.0)
blood pressure, JNC VI category
optimal BP (%) 6 (0.7) 48 (0.9) 28 (1.1)
normal BP (%) 9 (0.9) 21 (0.5) 19 (0.8)
high normal (%) 16 (1.2) 13 (0.4) 19 (0.9)
stage 1 hypertension (%) 41 (1.6) 13 (0.5) 24 (1.0)
stage 2 hypertension (%) 23 (1.4) 4 (0.2) 8 (0.5)
stage 3 hypertension (%) 5 (0.7) 1 (0.1) 2 (0.3)
left ventricular hypertrophy on 22 (1.6) 1 (0.2) 3 (0.4)
electrocardiogram (%)
Lifestyle factors
mean BMI (kg/m2) 27 (0.2) 26 (0.1) 28 (0.1)
obesity (% with BMI ⱖ30.0) 26 (1.4) 22 (0.7) 29 (1.0)
ever smoker (%) 61 (1.6) 53 (0.8) 63 (1.0)
current smoker (%) 15 (1.1) 28 (0.8) 28 (1.2)
physical activity (%) (ⱖ5 METS, ⱖ3 times/wk) 14 (1.1) 33 (1.1) 31 (1.2)
a
CVD, cardiovascular disease; NHANES, National Health and Nutrition Examination; SE, standard error; BMI, body mass index; JCN
VI, sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure; MET,
metabolic Equivalent Tasks.
b
NHANES III estimates were adjusted to the age (by decade), gender, race and prevalent atherosclerotic cardiovascular disease
distributions of the CHOICE cohort.

tion incorporates age, gender, total cholesterol, systolic BP, current reported in the USRDS (1). The proportion of those treated
smoking, LVH by EKG criteria, and glucose intolerance (defined as a with peritoneal dialysis is higher than USRDS because
diagnosis of diabetes, random glucose ⬎120 mg/dl, or urine dipstick CHOICE oversampled peritoneal dialysis patients. Diabetes
test positive for glucose) to estimate the 8-yr ASCVD risk in the
and hypertension accounted for approximately two-thirds of
Framingham cohort for those without a history of ASCVD (34,35). To
obtain 1- and 5-yr cardiovascular risk estimates, we assumed a con- ESRD, a figure similar to USRDS. However, the percentage of
stant risk over the 8 yr, converting the calculated 8-yr risk into a 1-yr ESRD attributed to hypertension was lower than in USRDS.
risk and 5-yr risk for each individual by means of the following CHOICE Study participants were somewhat healthier than the
formulas: national dialysis population, although for most conditions or
serologic factors, the difference was not great. The largest
1-yr risk ⫽ 关1 ⫺ e共0.125*共ln共1⫺8-yr risk兲兲兲兴
difference was for prevalent congestive heart failure (25% in
5-yr risk ⫽ 关1 ⫺ e共0.625*共ln共1⫺8-yr risk兲兲兲兴 CHOICE, 35% in USRDS).
Prevalence of ASCVD in the CHOICE cohort was deter-
Results mined for all ASCVD events (44%); MI (20%); MI or coronary
Patient Characteristics revascularization (32%); stroke (11%); stroke and carotid end-
Table 1 shows that the age, gender, race, and dialysis mo- arterectomy (17%); and peripheral vascular disease, including
dality distributions were similar to the US dialysis population bypass grafts, angioplasty, and amputation (26%).
1922 Journal of the American Society of Nephrology J Am Soc Nephrol 13: 1918–1927, 2002

Table 3. Comparison of lipid distribution adjusted to the CHOICE distribution of age, race, gender, and prevalent
cardiovascular disease, stratified by lipid-lowering medication usea
CHOICE Cohort NHANES III Population

On Lipid- Not on Lipid- Not on Lipid-Lowering

Lowering Medication Lowering Medication
Lipid Medication (n ⫽ 19,194)
(n ⫽ 169) (n ⫽ 872)

Unadjusted NHANES Estimates adjusted

Estimates (SE) Estimates (SE) III estimates (SE) to CHOICEa (SE)

On lipid-lowering medication (%) 16 (1.1) 2.8 (0.2) 8.4 (0.7)

Mean total cholesterol (mg/dl) 201 (4.1) 186 (1.7) 201 (0.7) 215 (1.3)
Total cholesterol, NCEP Category
⬍200 (normal) 57 (4.3) 65 (1.8) 52 (0.9) 38 (1.4)
200–239 (borderline) 20 (3.5) 23 (1.5) 30 (0.8) 35 (1.1)
ⱖ240 (high) 23 (3.7) 12 (1.2) 18 (0.6) 27 (1.3)
Mean LDL cholesterol (mg/dl) 111 (4.1) 106 (1.5) 126 (0.8) 136 (1.4)
High LDL cholesterol (% ⱖ160 mg/dl) 14 (3.2) 8.3 (1.1) 17 (0.8) 26 (1.7)
Mean apolipoprotein-B (mg/dl) 100 (2.8) 91 (1.1) 103 (0.8) 111 (1.0)
High apolipoprotein-B (% ⱖ140 mg/dl) 9 (2.5) 5 (0.8) 9 (0.6) 14 (1.1)
Mean HDL cholesterol (mg/dl) 43 (1.4) 43 (0.6) 51 (0.3) 50 (0.5)
Low HDL cholesterol (% ⬍40 mg/dl) 48 (4.4) 45 (1.8) 23 (0.8) 28 (1.1)
Total HDL:cholesterol ratio 5.2 (0.2) 4.7 (0.1) 4.3 (0.04) 4.8 (0.06)
Mean apolipoprotein-A1 (mg/dl) 138 (3.2) 131 (1.1) 143 (1.0) 144 (1.0)
Mean triglycerides (mg/dl) 250 (14) 189 (5) 139 (2.0) 162 (2.7)
High triglycerides (% ⱖ200 mg/dl) 52 (4.4) 34 (1.7) 17 (0.6) 23 (0.9)
High triglycerides (ⱖ200 mg/dl) and 30 (4.0) 24 (1.6) 9 (0.4) 13 (0.9)
low HDL (⬍40 mg/dl) (%)
a
NHANES, National Health and Nutrition Examination; SE, standard error.
b
NHANES III estimates were adjusted to the age (by decade), gender, race, and prevalent atherosclerotic cardiovascular disease
distributions of the CHOICE cohort.

ASCVD Risk Factors in the CHOICE Cohort cohort, most nonlipid ASCVD risk factors were still more
Table 2 lists the distribution of nonlipid ASCVD risk factors prevalent in CHOICE than the general population (Table 2). In
in CHOICE, compared with estimates from NHANES III. CHOICE, diabetes, hypertension, LVH by EKG criteria, and
Notably, the prevalence of diabetes, hypertension, and LVH by physical activity differed greatly in the direction of greater
EKG criteria was very high. Sixty-one percent of the cohort ASCVD risk, compared with the adjusted NHANES III esti-
reported previous smoking, compared with 15% for current mates. However, current smoking and obesity differed in the
smoking, and only 14% reported physical activity resulting in opposite direction, compared with the NHANES III adjusted
perspiration at a frequency of three or more times per week. estimates.
Table 3 presents the distribution of lipids, stratified by Table 3 also compares the lipid profile in CHOICE to the
lipid-lowering medication use. Sixteen percent of CHOICE unadjusted and adjusted estimates among NHANES III partic-
participants were taking lipid-lowering medications (including ipants not receiving lipid-lowering medications. CHOICE par-
HMG-CoA reductase inhibitors, fibric acids, nicotinic acid, or ticipants, regardless of lipid-lowering medication status, had
bile acid sequestrants). In the CHOICE cohort, total choles- lower total cholesterol, LDL cholesterol, apolipoprotein-B,
terol, LDL cholesterol, apolipoprotein-B, and triglycerides HDL cholesterol, and apolipoprotein-A1 levels and higher
were all significantly higher in those receiving compared with triglyceride levels, compared with the adjusted NHANES III
those not receiving lipid-lowering medication, and HDL cho- estimates.
lesterol levels were similar in the two groups. Most participants
not on lipid-lowering medications had normal or borderline Estimation of 1- and 5-yr ASCVD Risk
high total cholesterol levels, whereas 42% had either low HDL Table 4 presents the hypothetical 1- and 5-yr ASCVD risk
cholesterol or high triglycerides and 24% had both. for all those older than 40 without ASCVD (NHANES III did
not obtain EKG on those younger than 40 yr). Of the 459
Comparison between CHOICE and NHANES III CHOICE participants older than 40 without prevalent ASCVD,
After adjustment of the NHANES III prevalence estimates to complete information was available for 253 individuals, pri-
the age, race, gender, and ASCVD distribution of the CHOICE marily limited by the low number of individuals with an EKG
J Am Soc Nephrol 13: 1918–1927, 2002 Cardiovascular Risk Factors in ESRD 1923

Table 4. Distribution of Framingham risk factors and projected 8-year Framingham cardiovascular disease (CVD) risk
estimates among CHOICE and NHANES III participants ⬎40 years old without prevalent cardiovascular disease
CHOICE, ⬎40 years NHANES III, ⬎40 years
without CVD (n ⫽ 253) without CVD (n ⫽ 11,298)
Framingham Risk Factors and Risk Projections
Estimate (SE)a Estimate (SE)

Framingham risk factors

age (mean yr) 60 (0.8) 57 (0.4)
gender (% male) 49 (3.1) 45 (0.7)
mean total cholesterol (mg/dl) 191 (3.1) 217 (0.8)
mean systolic BP (mmHg) 151 (1.2) 129 (0.4)
current smoking (%) 15 (2.2) 23 (0.8)
left ventricular hypertrophy on EKG (%) 20 (2.5) 1 (0.2)
glucose intolerance (%) 68 (2.9) 12 (0.6)

1-year CVD risk projections (%)

mean overall risk in those ⬎40 years 3.1 (0.1) 1.4 (0.03)
by age group (yr)
40–49 (n ⫽ 67) 1.4 (0.2) 0.5 (0.01)
50–59 (n ⫽ 61) 2.8 (0.3) 1.3 (0.03)
60–69 (n ⫽ 67) 4.0 (0.3) 2.1 (0.03)
⬎70 (n ⫽ 58) 4.5 (0.3) 2.9 (0.04)
5-year CVD risk projections (%):
mean overall risk in those ⬎40 years 14 (0.6) 6.4 (0.1)
by age group (yr)
40–49 (n ⫽ 67) 6.5 (0.9) 2.3 (0.1)
50–59 (n ⫽ 61) 13 (1.1) 6.3 (0.2)
60–69 (n ⫽ 67) 18 (1.0) 10 (0.1)
⬎70 (n ⫽ 58) 20 (1.2) 13 (0.2)
a
SE, standard error.

available (n ⫽ 289). To test for bias because of the low (21,22), or local (23,24) studies of incident ESRD patients by
availability of EKG data, various factors were summarized in analyzing a wider range of ASCVD risk factors in a geograph-
those with and without an EKG, as follows: age (60.0 versus ically diverse and representative national cohort of incident
57.9 yr, P ⫽ 0.10), male gender (49 versus 49%, P ⫽ 0.93), dialysis patients and by making comparisons to the prevalence
systolic BP (153 versus 151 mmHg, P ⫽ 0.48), total choles- of risk factors in the general population.
terol (191 versus 191 mg/dl, P ⫽ 0.94), glucose intolerance (68
versus 62%, P ⫽ 0.29), current smoking (15 versus 20%; P ⫽
0.18), serum albumin (3.6 versus 3.6 g/dl; P ⫽ 0.94), hemat- Prevalence of Traditional ASCVD Risk Factors in
ocrit (32.1 versus 32.2%; P ⫽ 0.76), and serum creatinine (7.5 CHOICE
versus 7.6 mg/dl; P ⫽ 0.74). Furthermore, the projected 5-yr This study found a high prevalence for many traditional
ASCVD risk after excluding the LVH on EKG term was 11.1% ASCVD risk factors. The median age was high (60 yr), and
in those with an EKG compared with 10.5% in those without 54% of the participants were male. Diabetes, hypertension,
an EKG (P ⫽ 0.39), providing assurance that the two groups physical inactivity, hypertriglyceridemia, and low HDL cho-
are sufficiently similar to warrant use of the available data. lesterol levels were highly prevalent.
The CHOICE participants analyzed in the Framingham Overall, we found a higher prevalence of traditional ASCVD
equation analysis were slightly older than in NHANES III. risk factors in the CHOICE cohort than that reported by other
Total cholesterol and smoking were higher in NHANES; and national studies. Diabetes and smoking history were more
systolic BP, LVH on EKG, and diabetes were higher in prevalent in CHOICE than in the Case Mix Study (diabetes: 54
CHOICE. The 5-yr projected ASCVD risk was approximately versus approximately 40%; and ever-smokers: 61 versus ap-
twice as high in CHOICE (13%) as in NHANES (6%). The proximately 45%, respectively) (9). The Canadian study had
age-stratified comparisons show a greater relative difference in slightly higher current smoking rates compared with CHOICE
the younger than the older age decades. (22 versus 15%) but had a much lower diabetes prevalence (19
versus 54%) (4). Smoking history was also higher in CHOICE
Discussion than the 40% seen in DMMS Wave 2 Study (20). Both predi-
This cross-sectional study extends the ASCVD risk factor alysis-session mean systolic (149 mmHg) and diastolic (79
information reported by previous national (4,18,19), regional mmHg) BP were similar to DMMS Wave 2 Study (147 and 80
1924 Journal of the American Society of Nephrology J Am Soc Nephrol 13: 1918–1927, 2002

mmHg, respectively). Sixty-nine percent of the CHOICE co- cohort also stands in contrast to Culleton et al. (40), who
hort had a hypertensive predialysis BP. studied ASCVD risk factors among 664 individuals with mild
The prevalence of LVH on EKG (22%) was lower than chronic renal insufficiency (creatinine 136 to 265 ␮mol/L in
reported in the Case Mix Study (31%) but was similar that of men, 120 to 265 ␮mol/L in women). Although they found an
the DMMS Wave 2 Study (20%) (20). However, both the increased prevalence of diabetes (approximately 10%), AS-
DMMS Wave 2 Study and Case Mix Study included echocar- CVD (approximately 19%), hypertension (approximately
diographic data in the definition of LVH, whereas CHOICE 35%), and LVH on EKG (approximately 3.5%), relative to the
only used EKG criteria, which is known to underestimate the general population, the prevalence for all these conditions in
true prevalence of LVH. Foley et al. (36) found very high CHOICE is much higher. This suggests that ASCVD, hyper-
prevalence rates of LVH by echocardiogram (74%) in patients tension, diabetes and congestive heart failure either predispose
recruited within 1 yr of initiating dialysis. to the progression to ESRD, or are worsened by progression of
Only 14% of participants reported physical activity to per- chronic renal insufficiency, or are markers of a group of
spiration three or more times a week. This is consistent with individuals at high risk of progression. It is likely that all three
studies by Painter et al. (37,38) and Painter (39), which found processes play a role in progression to ESRD.
that ESRD patients have 63% of the exercise tolerance of
age-matched sedentary non-ESRD patients. Although physical
inactivity may contribute to the development of ESRD, it is Estimation of ASCVD Risk Attributable to Framingham
certain that the high degree of comorbidity associated with Risk Factors
ESRD itself promotes physical inactivity (the phenomenon of The very high prevalence of traditional risk factors in ESRD
reverse causality). The precise relationship between exercise may explain some of the excess ASCVD risk seen in ESRD,
and ESRD can only be determined by a prospective study of although it is unlikely to explain all of it (10). In an effort to
persons with chronic renal insufficiency. None of the other quantify ASCVD risk based on traditional risk factors alone,
studies of incident dialysis patients reported physical activity. Sarnak et al. (41) applied the Framingham risk equation (34) to
Total cholesterol and triglyceride levels in CHOICE were 1795 patients with chronic renal insufficiency. They found a
similar the DMMS Wave 2 Study (20), although we found weak negative correlation between the calculated ASCVD risk
lower total cholesterol and triglycerides in those not receiving and baseline GFR, suggesting that the Framingham risk factors
lipid-lowering medication and higher levels among those re- increase in prevalence as GFR declines. Cheung et al. (42),
ceiving lipid-lowering medication. The mean HDL cholesterol who also use the Framingham risk equation, report no signif-
level in CHOICE was much lower (43 mg/dl) than that re- icant difference between the calculated ASCVD risk among
ported in the DMMS Wave 2 Study (59 mg/dl) (20). prevalent ESRD patients in the Hemodialysis (HEMO) study
clinical trial compared with the general population, after age
Comparison with the General Population (NHANES adjustment.
III) In this analysis, we compared the Framingham risk equation
To our knowledge, no previous studies have attempted to score among those older than 40 without ASCVD in the
compare the ASCVD risk factor profile in incident ESRD NHANES population to similar individuals in CHOICE. The
patients with the general population. Direct comparisons are hypothetical 1- and 5-yr de novo ASCVD risk in the CHOICE
difficult to interpret because the age, race, gender, and, in cohort was approximately two times that of the NHANES III
particular, the prevalence of ASCVD differ greatly between the population. After age stratification, the relative difference be-
two populations. The higher ASCVD prevalence in ESRD tween the two groups was greatest in the youngest age groups.
inflates the prevalence of ASCVD risk factors, thus confound- It is important to stress that these calculated projections reflect
ing a direct comparison of ASCVD risk factors between the the estimated ASCVD risk that would result from this partic-
two populations. We therefore adjusted population estimates ular configuration of Framingham risk factors in the absence of
obtained from NHANES III to mirror the age, gender, race, and ESRD. They are not estimates of the true de novo ASCVD risk
ASCVD profiles of the CHOICE population (Table 2). among ESRD patients, for whom the actual ASCVD risk may
Many ASCVD risk factors were strikingly higher in be from 5 to 15 times higher.
CHOICE when compared with the adjusted NHANES III es- It may be inferred from these projections that as a group, the
timates, particularly diabetes, hypertension, LVH by EKG cri- Framingham risk factors explain some, but probably not all, of
teria, physical activity, low HDL cholesterol, and high triglyc- the extraordinarily high ASCVD risk seen in ESRD. Other
erides. These traditional risk factors have potential to explain studies of mortality in ESRD (9,14,16,43) have shown either
some of the increased ASCVD risk in ESRD. U-shaped or inverse relationships between mortality and vari-
The CHOICE estimates for current smoking, total choles- ous traditional risk factors such as BP and cholesterol— oppo-
terol, LDL cholesterol, and body mass index were lower in site to what is seen in the general population. Age, diabetes,
CHOICE than the adjusted NHANES III estimates, perhaps and LVH, however, are known to be risk factors for mortality
related to reverse causality (i.e., the comorbidity and malnu- in the ESRD population. Traditional ASCVD risk factors,
trition associated with ESRD may lead to lower cholesterol particularly cholesterol and hypertension, may interact with
levels and the decision to quit smoking, rather than vice versa). other nontraditional risk factors such as inflammation, comor-
The high prevalence of ASCVD risk factors in the CHOICE bidity, and malnutrition in the context of ESRD, thus altering
J Am Soc Nephrol 13: 1918–1927, 2002 Cardiovascular Risk Factors in ESRD 1925

their overall association with incident ASCVD in this Acknowledgments

population. We thank the patients, staff, and physicians who participated in the
CHOICE Study at Dialysis Clinic, Inc.; New England Medical Center;
New Haven CAPD; and Johns Hopkins University. We also acknowl-
Limitations of the Study edge and thank the CHOICE-DCI Clinical Liaison Committee (Thom-
Although we consider this a study of “incident” ESRD as Depner, M.D., H. Keith Johnson, M.D., K. Shashi Kant, Klemens
patients, the median time from initiation of dialysis to enroll- Meyer, M.D., Richard Sherman, M.D., Edward Schroeder, M.D.,
ment was 45 d. The study likely was not able to capture the Pradip Teradesai, M.D., John Van Stone, M.D., Alan Wasserstein,
very ill ESRD patients who die early after initiation of dialysis. M.D., Lucius Wright, M.D., Jackson Yium, M.D., and Philip Zager,
Every effort was made to include in the study all new dialysis M.D.) and the New Haven CAPD Clinical Liaison Committee (Fre-
patients at the 81 participating centers. However, Table 2 deric Finkelstein, M.D., and Alan Kliger, M.D.) for their support of
shows that CHOICE recruited somewhat healthier patients than and involvement in the CHOICE Study. We also recognize and thank
the USRDS population, suggesting that our data may underes- the CHOICE Scientific Advisory Committee (Paul Eggers, Ph.D.,
Sheldon Greenfield, M.D., H. Keith Johnson, M.D., Nathan Levin,
timate the true ASCVD risk factor prevalence in dialysis pa-
M.D., Robin Luke, M.D., Richard Rettig, Ph.D., and Paul Whelton,
tients. We estimate that this effect is small, however, because M.D., M.Sc.) for their scientific guidance and oversight of the
most differences between the two populations were minor. CHOICE Study. Presented in part at the American Society of Ne-
The comparisons between NHANES III and CHOICE data phrology Annual Meeting, October 2000. Published in part in abstract
must be interpreted with the understanding that some portion of form as Longenecker JC, Coresh J, Powe NR, Levey AS, Fink NE,
the differences in prevalence estimates between the two studies Klag MJ: Do traditional CVD risk factors explain race differences in
may be due simply to sampling design and the methods of data CVD risk and the excess CVD risk in ESRD? The CHOICE Study.
ascertainment, which were in some cases very different in the J Am Soc Nephrol 11:156A, 2000. CHOICE was supported by R01-
two studies. HS-08365 (Agency for Healthcare Quality and Research, formerly the
Another limitation is that the calculated risk estimates that Agency for Health Care Policy and Research) from June 1995 to May
use the Framingham equation in NHANES III are overesti- 2000 and is currently supported by the National Institute of Diabetes,
Digestive and Kidney Disorders, National Institutes of Health (NID-
mated to some degree as a result of the manner in which
DKD NIH; grant R01-DK-07024). Other NIDDKD NIH grants in-
ASCVD was defined. Included in the NHANES “no-ASCVD” clude the following: R29-DK-48362 (J.C.); K24-DK-02856 (M.J.K.);
group are those who have had a CABG, PTCA, carotid end- and K24-DK-02643 (N.R.P.). Support was also received from the
arterectomy or peripheral vascular disease, but who also never National Institute of Diabetes Digestive and Kidney Disorders, NIH
had an MI or cardiovascular accident. Such individuals, who (grant R01-DK-53869 to A.S.L.); and the National Heart, Lung and
were not excluded from the NHANES risk prediction analysis, Blood Institute, National Institutes of Health (grant K08-HL-03896 to
will inflate the ASCVD risk scores for the NHANES III J.C.L.). Lipid assays were supported by the National Center for
estimates in Table 4. However, we estimate that this effect is Research Resources (NIH grant M01-RR00052 to the Johns Hopkins
not large because the proportion of such individuals in the total University General Clinical Research Center).
NHANES III sample is low.
The CHOICE Study analyzed nonfasting specimens for the Appendix
lipid analyses. The results should be interpreted with the un- Derivation of Adjusted NHANES Weights
derstanding that the Friedewald equation, used to calculate The following equation was used to adjust the NHANES
LDL cholesterol levels, may underestimate the actual LDL weights to the age (by decade), race, gender, and ASCVD
cholesterol level. However, we also measured apolipopro- distribution of the CHOICE cohort:
tein-B, the primary lipoprotein in LDL cholesterol. The relative
ci
differences between CHOICE and NHANES are similar for
apolipoprotein-B and the calculated LDL cholesterol levels,
w new ⫽ wij ⫻
冘 wij
suggesting that any effect due to the use of nonfasting speci- j

mens is small.
Where wnew is NHANES population weight adjusted to the
CHOICE population by age decade, race, gender, and ASCVD
Summary status; wij is NHANES weight for each individual j in stratum
The prevalence of traditional ASCVD risk factors among i; ci is CHOICE proportion within each stratum i defined by
incident ESRD patients is very high. Even after adjustment for age decade, race, gender, and ASCVD status; and 兺wij is the
age, gender, race, and a high prevalence of ASCVD, most, but summation of the NHANES weights within each stratum i
not all, ASCVD risk factors are more prevalent in the ESRD defined by age decade, race, gender, and ASCVD status (the ci
population compared with the general population and may term is divided by this summation term such that the wnew
account for some of the increased ASCVD risk seen in ESRD. weights sum to 1.0).
Prospective studies in ESRD are needed to further define the
relationship between traditional ASCVD risk factors and inci- References
dent ASCVD, and clinical trials are needed to determine if 1. United States Renal Data System: USRDS 1999 Annual Data
reduction of risk factors will indeed decrease the incidence of Report. Bethesda, MD: National Institutes of Health, NIDDK,
ASCVD in ESRD. 1999
1926 Journal of the American Society of Nephrology J Am Soc Nephrol 13: 1918–1927, 2002

2. Parfrey PS, Foley RN, Harnett JD, Kent GM, Murray D, Barre 19. United States Renal Data System: IV. The USRDS Dialysis
PE: Outcome and risk factors of ischemic heart disease in Morbidity and Mortality Study: Wave 2. United States Renal
chronic uremia. Kidney Int 49: 1428 –1434, 1996 Data System. Am J Kidney Dis 30: S67–S85, 1997
3. Sarnak MJ, Levey AS: Cardiovascular disease and chronic renal 20. Stack AG, Bloembergen WE: Prevalence and clinical correlates
disease: A new paradigm. Am J Kidney Dis 35: S117–S131, 2000 of coronary artery disease among new dialysis patients in the
4. Churchill D, Taylor D, Cook R, LaPlante P, Barre P, Cartier P, United States: A cross-sectional study. J Am Soc Nephrol 12:
Fay W, Goldstein M, Jindal K, Mandin H, McKenzie J, Muir- 1516 –1523, 2001
head N, Parfrey P, Posen G, Slaughter D, Ulan R, Werb R: 21. Wolfe RA, Port FK, Hawthorne VM, Guire KE: A comparison of
Canadian hemodialysis morbidity study. Am J Kidney Dis 19: survival among dialytic therapies of choice: In-center hemodial-
214 –234, 1992 ysis versus continuous ambulatory peritoneal dialysis at home.
5. Rostand SG, Kirk KA, Rutsky EA: Relationship of coronary risk Am J Kidney Dis 15: 433– 440, 1990
factors to hemodialysis-associated ischemic heart disease. Kid- 22. Collins AJ, Ma JZ, Umen A, Keshaviah P: Urea index and other
ney Int 22: 304 –308, 1982 predictors of hemodialysis patient survival [published erratum
6. Lindner A, Charra B, Sherrard DJ, Scribner BH: Accelerated appears in Am J Kidney Dis 1994;24:157]. Am J Kidney Dis 23:
atherosclerosis in prolonged maintenance hemodialysis. N Engl 272–282, 1994
J Med 290: 697–701, 1974 23. Hylander B, Lundblad H, Kjellstrand CM: Changing patient
7. Herzog CA, Ma JZ, Collins AJ: Poor long-term survival after characteristics in chronic hemodialysis. Scand J Urol Nephrol
acute myocardial infarction among patients on long-term dialysis 25: 59 – 63, 1991
[see comments]. N Engl J Med 339: 799 – 805, 1998 24. Mailloux LU, Napolitano B, Bellucci AG, Mossey RT, Vernace
8. Parfrey PS, Harnett JD, Barre PE: The natural history of myo- MA, Wilkes BM: The impact of co-morbid risk factors at the
cardial disease in dialysis patients. J Am Soc Nephrol 2: 2–12, start of dialysis upon the survival of ESRD patients. ASAIO J 42:
1991 164 –169, 1996
9. Furth S, Hermann J, Powe N: Cardiovascular risk factors, co- 25. Powe N, Klag M, Sadler J, Anderson G, Bass E, Briggs W, Fink
morbidity and survival outcomes in black and white dialysis N, Levey A, Levin N, Meyer B: Choices for healthy outcomes in
patients. Semin Dial 11: 102–105, 1998 caring for end-stage renal disease. Semin Dial 9: 9 –11, 1996
10. Foley RN, Parfrey PS, Sarnac MJ: Clinical epidemiology of 26. Longenecker JC, Coresh J, Klag MJ, Levey AS, Martin AA, Fink
cardiovascular disease in chronic renal disease. Am J Kidney Dis NE, Powe NR: Validation of comorbid conditions on the end-
32: S112–S119, 1998 stage renal disease medical evidence report: The CHOICE
11. Coresh J, Longenecker JC, Miller ER, Young HJ, Klag MJ: study—Choices for Healthy Outcomes in Caring for ESRD. J Am
Epidemiology of cardiovascular risk factors in chronic renal Soc Nephrol 11: 520 –529, 2000
disease. J Am Soc Nephrol 9: S24 –S30, 1998 27. Paffenbarger RS, Jr., Blair SN, Lee IM, Hyde RT: Measurement
12. Levey AS: Controlling the epidemic of cardiovascular disease in of physical activity to assess health effects in free-living popu-
chronic renal disease: Where do we start? Am J Kidney Dis 32: lations. Med Sci Sports Exerc 25: 60 –70, 1993
S5–S13, 1998 28. Paffenbarger RS, Jr., Wing AL, Hyde RT: Physical activity as an
13. Special Report From the National Kidney Foundation Task Force index of heart attack risk in college alumni. Am J Epidemiol 108:
on Cardiovascular Disease: Controlling the epidemic of cardio- 161–175, 1978
vascular disease in chronic renal disease: What do we know? 29. Ainsworth BE, Haskell WL, Leon AS, Jacobs DR Jr, Montoye
What do we need to know? Where do we go from here? Am J HJ, Sallis JF, Paffenbarger RS Jr: Compendium of physical
Kidney Dis 32: S1–S199, 1998 activities: Classification of energy costs of human physical ac-
14. Lowrie E, Lew N: Death risk in hemodialysis patients: The tivities. Med Sci Sports Exerc 25: 71– 80, 1993
predictive value of commonly measured variables and an eval- 30. US Department of Health and Human Services: National Center
uation of death rate differences between facilities. Am J Kidney for Health Statistics, Third National Health and Nutrition Exam-
Dis 15: 458 – 482, 1990 ination Survey, 1988 –1994. NHANES III Laboratory Data File.
15. Zager PG, Nikolic J, Brown RH, Campbell MA, Hunt WC, CD-ROM series 11. Public use data file documentation no.
Peterson D, Van Stone J, Levey A, Meyer KB, Klag MJ, Johnson 76200. Hyattsville, MD: Centers for Disease Control and Pre-
HK, Clark E, Sadler JH, Teredesai P: “U” curve association of vention. Available from National Technical Information Service,
blood pressure and mortality in hemodialysis patients. Medical Springfield, VA, 1996
Directors of Dialysis Clinic, Inc. [published erratum appears in 31. US Department of Health and Human Services: National Center
Kidney Int 1998;54:1417]. Kidney Int 54: 561–569, 1998 for Health Statistics. Third National Health and Nutrition Exam-
16. Degoulet P, Legrain M, Reach I, Aime F, Devries C, Rojas P, ination Survey, 1988 –1994. NHANES III Examination Data
Jacobs C: Mortality risk factors in patients treated by chronic File. CD-ROM series 11. Public use data file documentation no.
hemodialysis: Report of the Diaphane Collaborative Study. 76200. Hyattsville, MD: Centers for Disease Control and Pre-
Nephron 31: 103–110, 1982 vention. Available from National Technical Information Service,
17. Parfrey PS, Griffiths SM, Harnett JD, Taylor R, King A, Hand J, Springfield, VA, 1996
Barre PE: Outcome of congestive heart failure, dilated cardio- 32. US Department of Health and Human Services: National Center
myopathy, hypertrophic hyperkinetic disease, and ischemic heart for Health Statistics. Third National Health and Nutrition Exam-
disease in dialysis patients. Am J Nephrol 10: 213–221, 1990 ination Survey, 1988 –1994. NHANES III Electrocardiography
18. United States Renal Data System: Special Study of Case Mix Data File. CD-ROM series 11. Public use data file documenta-
Severity IV: Comorbid conditions and correlations with mortal- tion no. 76200. Hyattsville, MD: Centers for Disease Control and
ity risk among 3,399 incident hemodialysis patients. Am J Kidney Prevention. Available from National Technical Information Ser-
Dis 20: 32–38, 1992 vice, Springfield, VA, 1996
J Am Soc Nephrol 13: 1918–1927, 2002 Cardiovascular Risk Factors in ESRD 1927

33. US Department of Health and Human Services: National Center 38. Painter P, Zimmerman SW: Exercise in end-stage renal disease.
for Health Statistics. Third National Health and Nutrition Exam- Am J Kidney Dis 7: 386 –394, 1986
ination Survey, 1988 –1994. NHANES III Household Adult Data 39. Painter P: The importance of exercise training in rehabilitation of
File. CD-ROM series 11. Public use data file documentation patients with end-stage renal disease. Am J Kidney Dis 24:
number 76200. Hyattsville, MD: Centers for Disease Control and S2–S9, 1994
Prevention. Available from National Technical Information Ser- 40. Culleton BF, Larson MG, Wilson PW, Evans JC, Parfrey PS, Levy D:
vice, Springfield, VA, 1996 Cardiovascular disease and mortality in a community-based cohort with
34. Kannel WB, McGee D, Gordon T: A general cardiovascular risk mild renal insufficiency. Kidney Int 56: 2214–2219, 1999
profile: The Framingham Study. Am J Cardiol 38: 46 –51, 1976 41. Sarnak MJ, Coronado BE, Greene T, Wang SR, Kusek JW,
35. Brittain E: Probability of coronary heart disease developing. Levey AS: Cardiovascular disease risk factors in chronic renal
West J Med 136: 86 – 89, 1982 insufficiency. Clin Nephrol 2002, in press
36. Foley RN, Parfrey PS, Harnett JD, Kent GM, Martin CJ, Murray 42. Cheung AK, Sarnak MJ, Yan G, Dwyer JT, Heyka RJ, Rocco
DC, Barre PE: Clinical and echocardiographic disease in patients MV, Teehan BP, Levey AS, and the Hemodialysis (HEMO)
starting end-stage renal disease therapy. Kidney Int 47: 186 –192, Study: Atherosclerotic cardiovascular disease risks in chronic
1995 hemodialysis patients. Kidney Int 58: 353–362, 2000
37. Painter P, Messer-Rehak D, Hanson P, Zimmerman SW, Glass 43. Avram MM, Mittman N, Bonomini L, Chattopadhyay J, Fein P:
NR: Exercise capacity in hemodialysis: CAPD and renal trans- Markers for survival in dialysis: A seven-year prospective study.
plant patients. Nephron 42: 47–51, 1986 Am J Kidney Dis 26: 209 –219, 1995

Access to UpToDate on-line is available for additional clinical information

at http://www.jasn.org/